5 Predictions for Design at the Edge of the AI Apocalypse

There's a version of this piece that functions mostly as therapy for anxious designers: a list of reasons AI might not be so bad after all, a few hopeful pivots, a reminder that human creativity cannot be replicated. This is not that piece.

What follows is an attempt to read the market honestly. Not the discourse, the market: job projections, capital flows, institutional gaps, and distribution structures. The AI story in design isn't primarily about creativity or its absence. It's about labour, ownership, and who ends up with the value.

1. AI is going to wipe out design jobs. But not the ones people are talking about.

The conversation about AI and design work tends to centre a particular designer: mid-career, working in branding or tech, anxious about their creative relevance. The data suggests the damage has been happening somewhere else entirely.

The World Economic Forum's 2025 Future of Jobs report — drawing on employer forecasts across 55 countries — ranked graphic design the eleventh fastest-declining occupation globally. Two years prior, the same report had it as moderately growing. Brookings researchers found that within eight months of ChatGPT's launch, freelance graphic design work had already contracted by 17 percent. A separate analysis tracked an 18 percent drop in job postings for graphic illustration.

These numbers are often discussed like forecasts. They are already describing the present. But the more useful framing isn't "AI is taking jobs." It's: which designers were already invisible?

The greeting card industry cut Hallmark's workforce from 22,000 to 10,500 employees between 2010 and 2015. Not from AI. From digitisation, from self-service platforms, from a market that decided templates were sufficient.

Some of these were graphic designers. They made things people actually used. They rarely appear in design industry discourse because that discourse is written largely by and for a different class of designer. The designers who lose first will be the ones who make wayfinding signs, retail packaging, promotional materials for businesses that couldn't afford an agency. Canva already proved their displacement was possible. AI is just faster.

2. Creative industry is not disappearing, but the labour market surrounding will.

The design industry has always needed luminous figures. Not for inspiration. For legitimacy. The celebrated studio, the institution-approved name, the designer who appears in monographs: these figures aren't just successful, they're structurally necessary. They justify the value of design as a category. Everyone further down the chain keeps working, for less, under their reflected light.

This is not changing. What's changing is how many people can exist in that shadow.

AI consolidates creative production upward, toward those with the capital to deploy compute and the infrastructure to operationalise it. The middle of the market compresses. The hierarchy stays intact with fewer people inside it. This is not a design-specific story. It's the same pattern across every sector automation touches: the top holds, the bottom falls, and what disappears is the range in between.

When the mythology of creativity-as-transcendence gets stripped away by market efficiency, what remains is a material record of what design labour produced and for whom.

3. Design quality will be paid for by the token. This is not a metaphor.

AI tools literally charge by the token — a precise description of how design value has always been distributed: by access to capital.

Better-funded companies always got better design. This was occasionally complicated by the passion economy: a junior designer who believed in a brief, who charged less than their skill warranted, who wanted the work more than the fee. A community publication could attract someone building a portfolio. The incentive structure was partly financial and partly something else.

That something else is being removed from the equation.

AI eliminates the volume work: the 400 mockups for a pitch, the twelve logo variants nobody asked for. This was grunt work. It was also training. It was how junior designers became senior designers, not by doing interesting work, but by doing repetitive work until interesting work became possible. Remove the grunt work and you do not simply automate the output. You sever the developmental pipeline that produced the labour pool in the first place.

The result isn't that everyone gets better design. It's that design quality stratifies more absolutely along capital lines than it ever did before. Companies with capital get compute and therefore quality. Companies without them get templates. That this was partly true before doesn't make the complete version of it an improvement.

4. Distributors and independent font designers will have to change their relationship.

The independent type designer's working model has depended on a bridge between creative labour and market access. That bridge was built on relationships: type directors who championed new work, connections made at conferences, distribution partnerships that provided modest but reliable income.

Algorithms do not maintain relationships. They reinforce whatever has already been validated at scale.

Bestsellers stay visible; everything else stays in the long tail. The person who remembered seeing your work at a conference, who pushed your typeface to a client out of genuine conviction: that function is being automated away, replaced by systems optimised for conversion, not discovery.

Meanwhile, the major distributors are under structural pressure to consolidate. Monotype, which controls a significant share of the global typeface market through its ownership of Linotype, Bitstream, FontShop, and MyFonts, was acquired by private equity firm HGGC for $825 million in 2019. By 2025, printers, agencies, and brand service providers were receiving surprise audit notices and five-figure invoices, demands for annualised licensing on fonts purchased outright years prior, and in at least one reported case, a renewal cost thirty times higher than what the licensee had been paying.

This is private equity logic applied to a creative infrastructure. A catalogue of 250,000 fonts is an asset to be extracted from. Independent designers whose work sits in that catalogue are not the bottom line; they're overhead.

The disparity is worth naming directly. As an example: looking from the outside, Monotype is aggressively extracting significantly more revenue from font licensing than it ever has from the same catalogues, built by the same designers. For those designers who haven't sold their catalogues outright and remain nominally independent, that increased revenue is not returning to them. The aggressive licensing enforcement is generating higher margins for the distributor, not the distributed.

The incentive is efficiency with what's already proven. Monotype just released an AI search tool and a standard for font discovery by agents. To think this will be used to help surface fonts that don't attend to their material bottom line is optimistic, but wrong. Monotype may be the first mover, but economic pressures will dictate that they're not the last.

The distribution partnership that once provided financial ballast is becoming untenable. The current model is structurally incapable of sustaining either side indefinitely; designers and distributors will have to choose: "me, or them?".

5. Creative education will discover that it was selling access, not knowledge.

The California College of the Arts, founded in 1907, is closing. San Francisco Art Institute is gone. Memphis College of Art. The Art Institutes system, which once enrolled tens of thousands of students annually, has completely shut down. Graduate arts enrolment dropped 9 percent in a single year.

This gets narrated as a funding crisis, a demographic problem, a liberal arts identity crisis. It is also a pricing problem that has been deferred for a long time. Australia made the mechanism explicit: in 2020, the federal government raised fees for arts and humanities students, and universities already weakened by the pandemic cut programs in response — animation, photography, printmaking, theatre, across ANU, Griffith, Monash, and others.

The prestige pipeline worked like this: admission to the right school conferred legitimacy; legitimacy conferred access to the right studios; the studios validated the credential; the schools pointed to those outcomes to justify their fees. The institution referred to itself to prove its own value. For a long time, the arrangement held because the industry agreed to believe in it.

And AI did not break this pipeline directly. The internet had already weakened the institution's monopoly on information; AI accelerated the collapse of its informational advantage.

Design history, visual systems, typographic theory, criticism: a working designer can now access most of this without institutional permission, and at a cost nowhere near $250,000. The experience of learning alongside other people, with mentors, inside a culture, still has real value. The institution's ability to charge for access to information that now exists on the open internet does not.

What's coming is not irrelevance. It's that the premium for the credential loses its justification. Designers who couldn't afford the room will compete directly with those who could. The institutional umbrella provided genuine protection for a long time. It is now priced above what the market will support.

The design industry keeps treating AI as a question of creativity because creativity is the last mythology it still knows how to defend. But the market has never organised itself around protecting creativity. It organises itself around reducing labour costs, consolidating infrastructure, and extracting value wherever production becomes easier to scale.

AI does not introduce a new logic into design. It intensifies an old one.

What disappears next will not be design itself. It will be many of the assumptions designers held about where value comes from, who gets to participate, and what institutions were actually protecting it in the first place.